福島原発災害以降、低線量放射線被ばくの問題が注目されている。これまで同様､放射線関連の公開データを用いて分析を行った。
･福島県甲状腺検査データ
福島県では現在4巡目の検査が行われているが､結果が公開されているのは1巡目(2011年10月-2014年3月実施､300,476名参加)､2巡目(2014年4月-2016年3月実施､270,516参加)である｡1巡目では甲状腺がん(疑いを含む) 116件の他に､直径5mm以下､同5.1mm以上の甲状腺結節がそれぞれ2,293件､1,244件見いだされた(それぞれ､参加者の0.039%, 0.76% and 0.41%)｡2巡目では甲状腺がんの検出数は大幅に減少すると想定されていたが71件の他､甲状腺結節についても各1,570件､2､219件が見いだされた(同0.026%, 0.60% ､0.80%)｡
公開されている市町村レベルのデータを用いて､これらを被説明変数､福島県健康調査･基本調査における外部被曝量(Akahaneetal.2013)､検査時年齢を説明変数としたポアソン回帰分析を行った｡59市区町村への2回の繰り返し測定と考えて､マルチレベル分析を適用し､潜在期間を考慮するために､被曝量と2巡目ダミーの交互作用を導入した｡
交互作用項は甲状腺がん(β= 0.631, z=1.91, P=0.056)､直径5mm以下の結節(β= 0.185, z=3.21, P=0.0013)､5.1mm以上の結節 (β= 0.167, z=1.97, P=0.049),のいずれについても有意となった｡この結果はこれらが放射線の影響によることを示唆する｡

･個人線量率関連論文のレビュー
放射線従業者の線量管理に用いられてきた個人線量計を用いることによって､住民の被曝量を予測もしくは管理しようという動きが見られる｡これに関して近年行われた研究についてレビューおよびNaito et al.(2016)データなどの再分析を行った｡その結果､｢サンプルの代表性の欠如｣｢分析の前提の確認不足｣｢航空機モニタリングの分解能や測定の信頼性の問題｣｢データの集計バイアスの可能性｣｢非効率的な分析｣｢統計手法に関する知識不足｣といった問題があることを指摘した｡このような分析に基づいて政策を決定すべきではない｡
Since Fukushima Nuclear Power Plant Disaster, health risk of radiation is a huge concern among Japanese citizen. Following previous studies, this research analyzed radiation related open data.
･Thyroid examination
Thyroid ultrasound examination (TUE) has been performed in subjects who were aged ≤18 years. In the 1st round TUE (October 2011 to March 2014), 116 malignancies (including suspicious cases, "malignancies", hereafter) were detected. Beside malignancies, 2,293 nodules with diameter ≤5 mm and 1,244 with nodules larger than 5mm were detected among 300,476 participants. They comprise 0.039%, 0.76% and 0.41% of participants respectively. In the 2nd round TUE (April 2014 to March 2016), while much less malignancy was expected, additional 71 malignancies, 1,570 nodules with diameter ≤5 mm and 2,219 with nodules larger than 5mm were detected among 270,529 participants as of June 2017. They comprise 0.026%, 0.6% and 0.8% of participants respectively. This study examined the relationship between radiation dose and the number of participants with thyroid nodules or malignancy, using publicly available municipality level data (N=59). To analyse two wave screening data, multi-level random-effect Poisson regression model was applied. The number of participants with thyroid abnormality was explained by log of external (effective) dose (Achene et al. 2013). Interaction between external dose and screening round dummy (0 for first and 1 for second screening) was also introduced to take into account latency.
Interaction terms between external dose and screening round dummy were positive and significant for small nodule (β= 0.185, z=3.21, P=0.0013), large nodule (β= 0.167, z=1.97, P=0.049), and thyroid malignancy (β= 0.631, z=1.91, P=0.056). These results are consistent with the conjecture that the first round TUE is the "baseline" that will not correlate with radiation level. The second round TUE detected thyroid cancer caused by radiation from Fukushima NPP that distributed heterogeneously among regions. Although this was an ecological study at the municipality level, health follow-up for children in Fukushima is urgent.

･Critical review of personal dosimeter related studies
Personal dosimeter that utilized to manage exposure of radiation workers has been applied to manage citizens' exposure in Fukushima. Through critical review and re-analysis of open data (Naito et al. 2016), the following limitations were identified: lack of representativeness of research objects, lack of reliability of measurement of location and personal dosimeter, neglecting of aggregation bias, inefficient analysis, lack of statistical method. Policymaking should not be based on these problematic studies.